高精度影像测量系统中图像的超分辨率重建  

Super-resolution reconstruction of image in high accuracy image measuring system

作　　者：张进[1] 王仲[1] 李雅洁[1] 叶声华[1] 

机构地区：[1]天津大学精密测试技术及仪器国家重点实验室,天津300072

出　　处：《光学精密工程》2011年第1期168-174,共7页Optics and Precision Engineering

基　　金：国家高档数控机床与基础制造装备重大科技专项(No.2009ZX04014-092)

摘　　要：为了提高影像测量系统的测量精度,研究了基于微位移错位的多幅图像超分辨重建技术。考虑制作成本和制作难度,提出了一种不苛求精确微位移的超分辨率重建技术。首先,获取随机微位移图像序列;然后,利用被测对象中易于实现高精度提取的特征点或采用模板衬底中的特征点计算出图像序列间的准确位移关系;最后,根据本文提出的数学模型重建出高分辨率图像并用于影像测量。实验结果显示,在图像分辨率提升至原始图像4倍时,对于不同大小的微位移,测量精度提升了29%~64%,表明该方法能够有效地重建出包含被测信息的高分辨率图像,进而提升影像测量系统的精度。To improve the measuring accuracy of an image measuring system,the super-resolution image reconstruction technology for an image sequence is researched based on the micro-displacement.As the high accurate micro-displacement is difficult to realize and very expensive,a super-resolution reconstruction method which doesn＇t require the high accurate micro-displacement is proposed.Firstly,the image sequence based on an arbitrary micro-displacement is obtained.Then,the accurate micro-displacement of image sequence is calculated by the feature points of a tested object or a template using a substrate.Finally,the high-resolution image is reconstructed by the mathematical model proposed in this paper for radiographic measurement.The experimental results indicate that when the image resolution is 4 times that of an original image,the measurement accuracy will promote 29% to 64% according to the different micro-displacements,which shows the method can reconstruct a high-resolution image including tested objects effectively,and can improve the accuracy of the system.

关 键 词：图像重建 超分辨率 影像测量 序列图像 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]